Behavior of Built-Up Shear Links Under Large Cyclic Displacement

ENGINEERING JOURNAL / FOURTH QUARTER / 2003 / 221

The towers of the Richmond-San Rafael Bridge (RSRB)play a significant role in the overall bridge seismic

behavior and performance. They are responsible for trans-ferring the inertia forces that are generated in the super-structure to the foundation during an earthquake.Therefore, the components of these towers lie in the directseismic load path and thus any premature failure in thesemembers will interrupt the load path causing inadequateseismic behavior.

The existing towers of the RSRB consisted of chevronbraced frames as shown in Figure 1. The chevron bracesconsist of built-up sections that are connected through dou-ble gusset plates. The design for seismic upgrading of thesetowers was prepared by the California Department ofTransportation (Caltrans) by the Gerwick, Severdrup,DMLM Joint Venture (Vincent, 1996). The seismic upgradeconsists of removing most of the existing chevron bracesand installing dual eccentric braced towers. Thus, the lat-eral forces are transferred through the new eccentric bracedtowers while the gravity load is transferred through theexisting towers legs. Figure 2 shows the elevation view ofthe proposed eccentrically braced tower that consists ofbuilt-up beams. During severe earthquakes, the linksbetween the eccentric braces become activated and dissi-pate the input energy by shear yielding. With this con-trolled yielding, the ultimate capacity and the maximumdrift can be quantified to a known level (Vincent, 1996).

Seismic resistant Eccentric Braced Frames (EBF) are lat-eral load resisting frames that are capable of providing highstiffness in the elastic range and high ductility in the inelas-tic range. During the 1980s, researchers (Malley andPopov, 1983; Hjelmstad and Popov, 1983; Kasai and Popov,1986; Engelhardt and Popov, 1989) investigated the cyclicbehavior of shear links and established design guidelinesthat are currently in many seismic provisions. More than

sixty shear links with various geometric configurations havebeen tested by Popov and others under large cyclic dis-placement (Itani, 1997). A common thread between allthese tests is the yielding of the web and the formation of a“ductile fuse.” This controlled yielding limits the frame’slateral strength to the ultimate capacity of the shear link.

The viability of EBF in seismic zones is a result of thesystem’s stable plastic rotation while maintaining the lateralstrength. Any premature local buckling or fracture in theelements of the shear link would cause an undesirable

Behavior of Built-Up Shear Links Under Large Cyclic DisplacementA.M. ITANI, S. ELFASS, and B.M. DOUGLAS

A.M. Itani is associate professor of civil engineering, Univer-sity of Nevada, Reno, NV.

S. Elfass, Bridge Research and Information Center Man-ager, department of civil engineering, University of Nevada,Reno, NV.

B.M. Douglas, emeritus professor of civil engineering, Uni-versity of Nevada, Reno, NV.

Fig. 1a. View of Richmond San Rafael Bridge.

Fig. 1b. View of Richmond San Rafael towers.

222 / ENGINEERING JOURNAL / FOURTH QUARTER / 2003

cyclic behavior and response. The design equations that arein the AISC Seismic Provisions (AISC, 1997) are tailoredtowards delaying local and global instabilities of the linkuntil a specified plastic rotation is achieved.

ISSUES RELATED TO BUILT-UP SHEAR LINKS

Figure 2 shows the elevation of the proposed new towers ofthe RSRB. These towers were designed to achieve a driftcapacity of 3 percent of their height (Vincent, 1996). Theproportions of the links along the height of the towers havebeen varied to ensure that they all reach the maximum rota-tion at approximately the same time. Therefore, built-uplinks with various proportions were utilized to control theultimate capacity of the tower. Changing the depth of thelink while maintaining the thickness of the web and thedimensions of the flanges, allowed for the capacity control.Structural steel for the flanges and the web for all the linksare intended to be taken from the same heat, thus achievinga common predictable yield stress (Vincent, 1996).

Built-up sections have an advantage over rolled shapesbecause they offer the flexibility of being proportioned toany desired dimensions. Thus, the designer can control thebehavior and the ultimate capacity of the section. Further-more, steel plates do not have significant variability in theyield stress since their production and rolling processes aresimpler than those of the rolled shape sections (Hamburgerand Frank, 1994).

OBJECTIVES AND SCOPE

The main objectives of this study were:

1. Investigate the cyclic behavior of built-up shear linksdesigned according to the 1997 AISC Seismic Provisions.

2. Collect experimental data on the plastic rotation and ulti-mate strength of built-up shear links.

3. Assess the performance of the eccentric brace-to-shearlink connections.

In order to achieve these objectives two full-scale shearlinks were tested under cyclic deformations. These linksrepresented the upper and lower bound of the web depth inthe proposed retrofit towers of the RSRB.

THE TEST SET-UP

The structural system of the test set-up was chosen to repli-cate the internal forces of one EBF bay. The sub-assemblywas selected to simulate the constraint conditions due to lat-eral loading on an EBF. Full-scale specimens were used toprevent introduction of scale effect on the response of thelinks. However, due to limitation of actuator load capacitya one-half EBF bay was used. To have the same internalforce effect in a one-half bay as in a full bay, a shear forcewas applied at the mid-length of the link and a roller sup-port was used at the connection between the beam and thecolumn. Figures 3 to 5 show one-half EBF bay under

Fig. 2. Elevation of proposed tower retrofit. Fig. 3. One-half EBF bay under applied force at link mid length.


applied load at link mid-length and the internal forces dueto the applied loading. As can be seen from these figures,the internal forces of the one-half bay are identical to theforces in a full EBF bay.

As indicated in Figure 3, the column is held in place,while the link beam is cycled back and forth. It is apparentthat the rigid-plastic kinematics of this subassembly is iden-tical to the one of a full EBF bay.

A schematic representation of the test set-up is shown inFigure 6. The loading unit is composed of double actuators,loading unit, hydraulic pump, and a controller. Two actua-tors were combined together to apply the cyclic displace-ment. A loading unit was used to combine the two actuatorsand apply the load to the link. The push-pull capacity of thetwo actuators combined is ±750 kips and ±10 in.

TEST SPECIMENS

The links were designed according to the AISC SeismicProvisions (AISC, 1997). The length of the link was cho-sen to be less than 1.6Mp /Vp. Therefore, the inelasticbehavior will be dominated by shear yielding. The flangesand the web of the link complied with the width-thicknessratios that are specified in the seismic provisions. The spac-ing of the intermediate web stiffeners was based on theAISC specified value of 30tw−d/5 to achieve a plastic rota-tion equal to 0.08 radians.

Two built-up specimens were tested in the experimentalprogram. Specimen 1, BU30, consisted of a built-up sec-tion that has a total depth of 30 in. The flanges of BU30were 14 in. × 1½ in. while the web was 27 in. × 3/8 in. Thetotal length of the beam was equal to 17½ ft and the linkportion was equal to 6 ft measured from the pin to the faceof the diagonal stiffener as shown in Figure 7.

The flanges were welded to the web using a 5/16-in. fil-let weld on the two sides of the web. Two 123/8 in. × 1 in.continuity plates were added to the built-up section at theintersection with the eccentric brace. These plates werewelded to the web and to the top flange using 3/8-in. fillet

Fig. 4. Normal force diagram under applied force at link mid length.

Fig. 5. Shear force diagram under applied force at link mid length. Fig. 6. Schematic view of the test set-up.


The CJP groove weld of the continuity plates to thebottom flange and the CJP weld of the web infill plateto the bottom flange shall be complete prior to weld-ing the web and web infill plate to the continuity plate.The fillet weld attaching the continuity plates and theweb infill plate to the underside of the top flange shallbe the last weld made. The bottom end of the eccentric brace was bolted to a

stub member as shown in Figure 6 in order to remove andreplace the test specimens.

The second specimen, BU16, consisted of a built-upbeam with a total depth of 16 in. Figure 8 shows the dimen-sions and details of BU16.

Lateral support was provided at the top and bottomflanges at the end of the link next to the eccentric brace con-nection. A W10×68 section was used at the top and bottomflange of the link as a lateral support as shown in Figure 9.

MATERIAL PROPERTIES

Caltrans Standard Specifications (Caltrans, 92) and RSRBSpecial Provisions were used to specify materials and thefabrication process. The material section in the RSRB Spe-cial Provisions stated:

welds and a complete-joint-penetration groove (CJP) weldto the bottom flange. A 1-in. thick web doubler plate wasadded to the panel zone at the intersection with the eccen-tric brace. Coped holes were used at the four corners of thebeam panel zone with a radius equal to 1½ in.

Twelve 6 in. × 3/8 in. stiffeners were mounted on each faceof the link. The first stiffener was spaced at 3¾ in. from thecontinuity plate and the rest were spaced at 5¼ in. based onthe equation in the AISC Seismic Provisions. Figure 7shows the details of the test specimen. The intermediatestiffeners were fillet welded to the web and the flanges ofthe shear link using ¼-in. fillet weld. Three additional stiff-eners, 6 in. × 3/8 in., were added outside the link on one faceof the web as shown in Figure 7. These additional interme-diate stiffeners (outside the shear link) were designed andspaced according to the AASHTO Standard Specifications(AASHTO, 1996) to prevent any web buckling due to thepresence of the axial force and large bending moment inthat region.

The eccentric brace, W12×190, was welded to the bottomflange using complete-joint-penetration groove welds. Thewelding sequence was specified on the structural drawingsas:

Fig. 7. Details of BU30.

The material for the link beam member flanges andweb shall conform to ASTM Designation:A709/A709M Grade 50T2. Ultrasonic testing shallconform to the provisions of ASTM Designation:A578/A578M. Material shall conform to acceptancestandard level III. In addition, the maximum yieldstrength for the material used for the link beam mem-ber flanges and webs shall have maximum yieldstrength no greater than 60 ksi.Six 18-in. long ASTM A6 coupons with gauge length of

8 in. × 3/8 in. were tested using Instron 4210 with hydraulicgrips and laser extensometer. To trace the stress-strain curveof the web, coupons were removed from the following weblocations:

• Parallel to direction of rolling, specimens L1 and L2


• Inclination of 45° to direction of rolling, specimens I1and I2

• Perpendicular to direction of rolling, specimens P1 andP2

The L specimens were oriented along the longitudinaldirection of the shear link. The average yield stress andultimate strength of coupons L1, L2, P1, and P2 is equal to57 ksi and 81 ksi, respectively. It is interesting to note herethat L1 and L2 coupons have the highest yield stress, whileP1 and P2 coupons have the lowest yield stress. Also, thefracture strain for L and P coupons was almost identical,while the fracture strain for I coupons was 30 percent lessthan the fracture strain of L and P coupons. Table 1 lists thetest results of these coupons. The average yield stress and

Fig. 8. Details of BU16.

Coupon Fya (ksi) Fu (ksi) Fracture Strain (in./in.)

L1 59 82 19 L2 60 82 20 P1 54 80 21 P2 54 78 20 I1 58 80 14 I2 59 81 11

Table 1. Mechanical Properties of Web Material Using ASTM A6 8-in.-Gauge Coupons

ultimate strength of the flanges of the shear link were equalto 54 ksi and 84 ksi, respectively.

TEST SPECIMEN FABRICATION

Christie Constructors, Inc., Richmond, CA fabricated thetest specimens. The fabrication was performed according toCaltrans Standard Specifications and AASHTO/AWS D1.5Provisions. The fabricator developed the Welding Proce-dure Specifications (WPS) that were reviewed andapproved by Caltrans representatives. A Certified WeldingInspector (CWI), as per the AWS-QC-1 standards, fromTesting Engineers, Inc., Oakland, CA inspected andapproved the fabrication of the test specimens according toAWS D1.5 requirements.


Gas metal arc welding (GMAW) was used to weld thetest specimens. The electrodes for GMAW conformed tothe requirements of AWS A5.18 Specification for CarbonSteel Filler Metals for Gas Shielded Arc Welding with clas-sification ER 70S-2. All welders that worked on the testspecimens were qualified by the tests described in AWSD1.5 and their qualifications were verified by the CWI. Thewelding parameters including volts, amps, wire feed speed,and arc travel speed, were within the recommended rangesset by electrode manufacturer from the approved AWSspecifications and classifications.

The minimum pre-heat and inter-pass temperatures formembers up to ¾ in. is 50 °F, over ¾ in. thru 1½ in. is 100 °F,and over 1½ in. through 2½ in. is 200 °F. All weld passeswere made using stringer beads with no weaving or washpasses. Except for the root pass, passes did not connect thesides of the weld groove. Also, except for the root and thefinal passes, peening was performed to each pass when theweld is at a temperature of 150-500 °F. Peening was per-formed with a slag gun at right angles to the weld makingfive to six passes along the length of the weld and using adull tool to prevent sharp notches and cuts. Stress reliefcope holes in the webs were used to allow the weld joint andback-up plates to be continuous. Stress relief copes wereground to remove all defects including notches. Back-platesand run-off plates were removed and cleaned according tothe AWS D1.5 Specifications.

INSTRUMENTATION

Strain gauges and rosettes were mounted on various loca-tions of each test specimen as shown in Figure 10. Cableextension transducers were used to measure the displace-ment profile along the beam as shown in Figure 11. Allthese devices were connected to a data acquisition systemwith a sampling rate of two readings per second.

Fig. 10. Location of strain gauge and rosettas in BU 30. Fig. 11. Location of wire potentiometer.

Fig. 9. Schematic view of the lateral support.

CAPACITY OF TEST SPECIMENS

Some of the proportions of the two built-up shear links felloutside the ranges of previously tested shear links. Kasaiand Popov (1986) collected data of thirty shear links thatwere tested by Popov and others. Table 2 shows a compari-son between the proportions of the shear links that werereported by Kasai and the two test specimens. As can beseen from this table, the built-up shear links are signifi-cantly different from the previously tested shear linksspecifically in the values of tf /tw, D/tw, a/tw and a/D.

FLEXURAL AND SHEAR STRENGTH

The satisfactory cyclic performance of shear links dependsmainly on the ratio between shear and flexural capacity.The yield mechanism for the link can be identified as fol-lows:

• Development of plastic hinges at the ends of the link.

• Development of plastic hinges at the ends of the link inthe presence of high shear.

• Development of shear yielding accompanied with yield-ing at the ends of the link.

The expected plastic moment capacity of the section isequal to

Mpe = Z Fye

where Fye = the expected yield stress based on the AISC Seis-

mic Provisions (AISC, 1997)Fye = Ry Fy

The value of Ry for steel plate is equal to 1.1 for A572Grade 50 steel.

The plastic shear capacity is equal to:

Vp = 0.6 Fy (d−2tf) tw

According to the AISC Seismic Provisions, the nominalshear strength of the link is equal to the lesser of Vp or 2 Mp /e,where e is the link length. Table 3 presents the nominalshear of the two test specimens, BU30 and BU16. Based onthis table it is clear that the test specimens will be domi-nated by web yielding since e is less than 1.6Mp /Vp.

Table 4 shows the difference between the expected yieldshear strength (Vpe) and the actual shear strength (Vpa). As


Ratio

Previous Test Ranges

(Kasai and Popov, 1986)

BU30

BU16

tf /tw 1.2-1.67 4 4

b/2tf 5.4-9.6 4.67 4.67

D/tw 40-57 72 34.7

a/tw 21-57 14 21.3

a/D 0.42-2.4 0.19 0.62

Table 2. Geometric Ratios of Previously Tested Links and the Built-Up Specimens

Specimen

Mp (k-ft)

Vp (kips)

e (ft)

1.6Mp /Vp

(ft)

2Mp /e (kips)

Vn (kips)

BU30 2,779 304 12 15.3 463 304 BU16 1,335 146 12 16.7 223 146

Table 3. Shear Capacity of Test Specimens Using Nominal Yield Stress

Specimen Vp Vpe Vpa Ry Rya

BU30 304 334 353 1.1 1.16 BU16 146 161 170 1.1 1.16

Table 4. Nominal, Expected and Actual Shear Capacity of Test Specimens

can be seen from this table, the value of Ry recommendedby AISC for A572 Grade 50 steel underestimates the actualcapacity by almost 6 percent.

PLASTIC ROTATION

According to AISC Seismic Provisions (AISC, 1997), theplastic rotation for links of lengths 1.6 Mp /Vp or less withintermediate web stiffeners spaced at intervals not exceed-ing (30tw-d/5) should be at least 0.08 radians. The plasticrotation is defined as the inelastic angle between the linkand the beam outside of the link. For BU30 and BU16 thevalue of 1.6 Mp /Vp is equal 15.3 ft and 16.7 ft, respectively,which exceed the specified 12 ft link length. In addition, theweb stiffener spacing for the two shear links satisfied thespecified spacing in the AISC requirements.

LOADING HISTORY

The loading history was specified according to Guidelinefor Cyclic Seismic Testing of Components of Steel Struc-

tures (ATC, 1992). The test protocol is shown in Figure 12.The yield displacement, δy, was determined experimentallyfor each BU30 and BU16 specimens.

EXPERIMENTAL TESTING

Specimen BU30

Figure 13 shows the response of specimen BU30. The hor-izontal axis is the applied displacement while the verticalaxis is the shear force resisted by the specimen. The shearforce is the combination of the forces measured from thetwo actuators, while the applied displacement is measuredform the wire potentiometer C1. At an applied displacementequal to 1.2 in., the web started to yield. Inclined yield lineinitiated in most web panels at a load equal to 348 kips. The1.2 in. displacement was specified as the yield displace-ment, δy, of BU30. The specimen was then subjected to thedisplacement cycles according to the loading history, caus-


Fig. 13. Force-displacement response of BU30.

Fig. 14. Bottom flange fracture of BU30. Fig. 15. Force-displacement curve of BU16.

Fig. 12. Loading history.

ing web yielding to become more pronounced. As theapplied displacement increased to 4δy and 5δy, yieldingstarted to spread outside the web of the link. Yield lines ini-tiated at the top and the bottom flanges of the ends of theshear link. At a displacement of 5δy excursion, yielding wasobserved along the top and bottom link flanges at the edgeof the pin plates where the load was applied.

The specimen was then subjected to two cycles at the 6δy

displacement level with the applied load reaching 600 kips.The web outside the shear link started to yield in addition tomore yielding in the top flange. The displacement was thenincreased to reach 7.68 in., 6.4δy, representing 8 percentplastic rotation with a corresponding load of 633 kips.

Upon completion of the first cycle at 6.4δy and proceed-ing to the push excursion of the second cycle, a sudden frac-ture occurred in the bottom flange of the shear link next tothe eccentric braced weld. The bottom flange of the shearlink was completely fractured as shown in Figure 14. Thecrack then propagated through the web of the first panel toalmost one-third of the web depth. After the fracture, theload dropped to almost zero and the test was stopped.

Specimen BU16

Figure 15 shows the response of specimen BU16 with thevertical and horizontal axes similar to Figure 13. At a dis-placement equal to 1.5 in., the web started to yield. Yieldline initiated in most web panels at a load equal to 185 kips.The 1.5 in. displacement was specified as the yield dis-placement of BU16. The specimen was subjected to cyclesaccording to the loading history. As the displacementincreased to 4δy and 5δy yielding started to spread along thebottom and top flanges of the link similar to specimenBU30.

The specimen was then subjected to two cycles at 6δy dis-placement level with the applied load reaching 325 kips.Upon completion of the first cycle at 7δy and proceeding to

the pull excursion of the second cycle, a fracture occurredin the third and fourth web panel as shown in Figure 16.Severe flexural distortion at the flanges of the shear linkoccurred at the fractured web location. The load capacitydeteriorated and reached almost 200 kips. At that level, thetest was stopped. After the test, a hairline crack was noticedat the welded connection between the bottom flange andeccentric brace connection.

FRACTURE AND MATERIAL EVALUATION OF BU30 SPECIMEN

The two test specimens showed that the connection betweenthe eccentric brace and the bottom flange of the shear linkwas susceptible to fracture at high strain levels. SpecimenBU30 suffered a complete fracture in the bottom flangewhile a hairline crack was noticed in Specimen BU16.Schwein/Christensen Laboratory, Lafayette, California,evaluated the connection between BU30 and the eccentricbrace. The purpose of this evaluation was to comment onthe fracture mode and quality of materials used to fabricatethe welded assembly.

Fractographic Examination

The fracture occurred in the lower flange, immediatelyadjacent to the complete-joint-penetration groove weld jointat the eccentric diagonal intersection. The fracture propa-gated completely through the bottom flange of the shearlink. It arrested after it propagated approximately 6 in. inthe web.

The fracture surfaces were examined visually, and withthe aid of optical scanning electron microscopes. The frac-ture features observed indicated a mixed mode of failure.Ductile shear rupture and brittle cleavage zones were bothobserved in the fracture zone. The fracture features weresmeared/crushed along the center-bottom edge of theflange, which may be attributed to compressive load cyclingafter initial cracking occurred in the bottom flange.

Chevron features on the fracture surface indicate the brit-tle cleavage zone of fracture initiated near the mid-thick-ness of the flange directly below the web-cope area.Immediately above and below this point is a zone of ductileshear rupture. It is significant to note that the brittle cleav-age fracture did not initiate at a weld defect.

With respect to weld quality, a slight overlap defect wasobserved at the toe of the weld that connects the web to thelink bottom flange at the cope area. Also it appeared thatthe web cope weld access hole was flame cut without finishgrinding. These workmanship flaws do not appear to havepresented a significant stress riser to act as an initiation sitefor brittle fracture.

Link bottom flange thickness measurements indicate areduction in cross-sectional thickness of up to 7 percent at


Fig. 16. Web fracture of BU16.

the fracture. This reduction in thickness is an indirect indi-cation of some limited plastic deformation in the beam priorto fracture.

Microstructural Examination

A longitudinal cross section was metallurgically preparedfrom the link flange at the fracture. This included a portionof the heat-affected zone (HAZ) from the CJP weld. A fer-ritic-pearlitic microstructure was observed in the base metalof the beam flange. This is a normal structure for ASTMA709 Grade 50 steel. The HAZ showed a mixed ferrite-pearlite-banite structure that is commonly observed withnon-equilibrium cooling associated with welding.

Mechanical Properties

Base Metal Tensile Tests

Longitudinal tensile coupons were machined from theflanges (at ¼ in. thickness) of the link and the diagonalbrace. Both samples were removed from a location withinapproximately 6 in. of weld fracture area. Additionally, a

transverse tensile coupon was machined from the beamweb. The samples were tensile tested in accordance withASTM A370.

The results are shown in Table 5. It is interesting to notehere that the sample coupon from the shear link web exhib-ited a very high yield stress and yield-to-tensile ratio. Addi-tionally, there was no lower yield point observed and thepercent elongation was equal to 15 percent. From theresults, it can be concluded that the web material was sub-stantially strain hardened as a result of the cyclic testing.Although not as pronounced, a similar trend showing rela-tively high yield strength was observed in the shear link bot-tom flange indicating some strain hardening.

Notch Toughness: Base Metal and Weld Metal

Charpy-V-Notch (CVN) impact specimens were machinedfrom the flanges of the beam and diagonal brace. A set ofCVN specimens were also machined from the CJP weldnear the fracture, that joins the diagonal brace to the bottomflange of the shear link. The CVN specimens were alltested at room temperature (73 °F) in accordance to ASTM


Sample Location

Absorbed Energy (ft-lb)

95.5 94.3

108.9

Bottom Flange Average =99.6

119.2 95.2

113.6

Diagonal Brace Bottom Flange

Average=109 109.6 107.5 119.4

Weld Metal

Average =112

Table 6. Charpy V-Notch Toughness Tested at Room Temperature

Sample

BU30 Bottom

Flange

BU30 Web

Diagonal Brace Bottom Flange

Required A572

Grade 50 Yield Strength

(ksi)1

N/A N/A 58 50

Yield Strength

(ksi)2

69 97 52

Tensile Strength (ksi)

86 102 77 65

Elongation in 2 in. 27% 15% 33 21%

Table 5. Mechanical Properties for BU30 Specimen After Cyclic Testing

1 Yield strength determined at proportional limit (upper yield point)2 Yield strength determined at 0.2% offset method (lower yield point)

A370. The impact test results are shown in Table 6. Allthree materials (in other words, bottom flange, diagonalbrace and CJP weld) exhibited relatively high notch tough-ness.

Microhardness Test

A Knoop microhardness survey was conducted on the sam-ple previously prepared during the microstructural exami-nation. The hardness was tested in the lower flange basemetal and HAZ of the shear link near fracture. The hard-ness results are shown in Table 7. The average base metaland HAZ hardness was Rockwell B97 and C34, respec-tively. The base metal hardness is somewhat higher thannormal A709 Grade 50 material and is probably due to thework hardening from the cyclic testing. The HAZ hardnessindicates some partial banitic phase transformation fromnon-equilibrium cooling. While the HAZ hardness is on thehigh side, the levels are below that which is normally asso-ciated with hydrogen cracking or embrittlement.

Discussion of Fracture Inspection

The results of the fracture evaluation of a BU30 connectionindicated that the materials and workmanship were gener-ally in compliance with the appropriate standards and spec-ifications, with the exception of the slight flaws previouslynoted at the cope/weld access hole. The extension of the fil-let weld into the weld access hole may have contributed tosome reduction in plastic rotation by focusing additionalstress concentration and localized strain demand in theflange/cope area. The degree to which this flaw diminishedthe performance is uncertain.

In examining the fracture surface it is evident that thereis a significant stress/strain gradient horizontally across andvertically through the shear link bottom flange with the

maximum stress concentration just below the web/cope areaadjacent to the CJP weld. This stress concentration createsa substantial demand for localized strain/ductility. It isapparent that the connection exhibited some initial ductilityas evidenced by the areas on the surface that show ductileshear rupture. Once the demand ductility increased, largetri-axial stresses developed from strain hardening near theweb/cope area that resisted further plastic flow, ultimatelycausing the final brittle cleavage fracture.

ANALYSIS OF EXPERIMENTAL RESULTS LINKYIELD AND ULTIMATE SHEAR STRENGTH

Link Yield Strength

The yield strength of shear links provides an importantbenchmark during the design process of EBFs. Lateralloading on an EBF, causes a constant shear along the lengthof the shear link and opposing bending moments at the endsof the link. Therefore, the shear link will be subjected tohigh shear and high bending moments at its ends. Accord-ing to perfect plasticity, the presence of the shear and bend-ing moment cause a reduction in both shear strength andflexural strength. This reduction can be predicted analyti-cally using moment-shear (M-V) interaction surfaces.However, experimental data of short shear links conductedby Kasai and Popov (1986) and Engelhardt and Popov(1989) showed that M-V interaction does not exist at theyield limit state. Therefore, the yield strength can be takenas Vp and the flexural strength can be taken as Mp regardlessof the magnitude of shear force.

Table 8 presents the values of the shear yield strengthbased on the specified (Vp), expected (Vpe), actual (Vpa), andmeasured (Vpm) values for BU30 and BU16. Table 8 alsoshows the ratios Vpm to Vp and Vpm to Vpa. As can be seen


Sample Location Knoop Microhardness Rockwell Hardness Approximate Tensile Strength 1 (ksi)

232 B96 103 240 B98 109

Beam Bottom Flange Base Metal Near Fracture 243 B98 109

345 C34 153 355 C35 157

Heat Affected Zone

338 C33 148

Table 7. Microhardness Test Results of BU30 Connection

1 Rockwell hardness and approximate tensile strength values are from published conversion tables.

Analytical Experimental Yield Overstrength

Specimen Vp Vpe Vpa Vpm Vpm /Vp Vpm /Vpa

BU30 304 334 353 348 1.15 0.99 BU16 146 161 170 185 1.27 1.09

Table 8. Shear Yield Strength According to Analytical and Experimental Values and Yield Overstrength Ratios

from this table the specified shear strength underestimatedthe value by 15 percent and 27 percent for BU30 and BU16,respectively. On the other hand, using the actual value forthe web yield stress was able to predict the measured valueswithin 1 percent as Table 8 shows. Therefore, it appears thatneglecting the M-V interaction and using the actual webyield stress provides a reasonable estimate of the shear yieldstrength for the tested built-up specimens as in the case ofrolled shape short shear links.

Link Ultimate Shear Strength

The basic design philosophy of EBF is to limit the inelasticactivity to the link, while all other members, outside thelink, stay in the elastic range. Following capacity designprocedures, the members outside the link must be designedfor the maximum forces generated by the link. Therefore,developing a reasonable overstrength ratio of the link shearforce is an important issue in EBF design. Underestimatingthe maximum shear force that a shear link can generate maylead to a non-ductile behavior such as buckling of theeccentric brace or beam failure outside the link. The com-mentary of the 1997 AISC Seismic Provisions states:

The design strength of the diagonal brace is requiredto exceed the forces corresponding to Ry times thenominal shear strength increased 25 percent for strainhardening. That is with φ equal to 0.85 for axial com-pression in the brace, the effective overstrength factor(assuming Ry=1.1) becomes 1.25(1.1)/0.85 or about,1.6 for steel with low variability in Fy and (assumingRy=1.5) about 2.2 for steel with high variability. Withφ equal to 0.9 for flexure in the beam or diagonalbrace, the effective overstrength factor becomes1,25(1.1)/0.9, or about 1.5, which represent a slightrelaxation from the test criterion for steel low vari-ability.The commentary also stated that: “The overstrength fac-

tor was developed from tests on typical beams with usualflange thicknesses. For link beam with relatively thickflanges, this factor may need to be increased.”

Table 8 presents the values of the ultimate shear strengthof the test specimens. The overstrength ratio (Ω) for BU30and BU16 can be determined by dividing the measured ulti-

mate shear strength to yield shear strength. Table 9 showsseveral over strength ratios based on the measured ultimateshear strength to Vp, Vpe, Vpa, and Vpm. As can be seen fromthis table, Ωs was equal to 2.08 and 2.12 for BU30 andBU16, respectively.

As mentioned earlier, link overstrength can be attributedprimarily to strain hardening and to the actual yieldstrength. Using Ωa as an indicator for the strain hardeningratio, it shows that the two test specimens were subjected tosignificant strain hardening which can be estimated as 1.79and 1.82 for BU30 and BU16, respectively.

It is important to note here, that the failure mode of BU30was the fracture of the shear link bottom flange. Therefore,the force 633 kips at which the bottom flange of the shearlink fracture does not represent the actual ultimate shearstrength. Based on this investigation, it can be determinedthat the overstrength factor for BU30 and BU16 is equalto 2.10.

Link End Moment

The magnitude of the link end moment is an importantparameter because it dictates the design of the diagonalbrace and the beam outside the link. The maximum bend-ing at the end of the shear link (Vu × L) for BU30 and BU16is equal to 3,798 k-ft and 1,860 k-ft, respectively. Thesevalues exceed the Mp for each section by 15 percent and 23percent for BU30 and BU16, respectively. As discussedpreviously, the ultimate strength of these built-up links fordesign purposes has been taken as 2.1 times the yieldstrength based on simple plastic theory with no M-V inter-action. This approach would predict the ultimate endmoment of 2.1Mp for these links. This indicates that theoverstrength 2.1 is high for the flexural design of the beamsegment outside the shear link but may be used for thedesign of the axial forces in the beam segment and theeccentric brace. More experimental tests on built-up shearlinks will shed more light on this important issue.

Moment-Rotation Relationships at Link Ends

Rotations at the end of the link were determined based onthe reading of the wire potentiometers C1 and C6. The rota-


Specimen Ωs Ωe Ωa Ωm

BU30 2.08 1.90 1.79 1.82

BU16 2.12 1.93 1.82 1.68

Table 9. Overstrength Ratios For Shear Ultimate Strength

tion, γ, at the end of the shear link is defined by the relativeend deflection of the shear link divided by the link length.The rotation is equal to:

where δ1 and δ6 is the displacement of wire potentiometerof C1 and C6, and L is the length of the shear link as shownin Figure 17. Figure 18 shows the displacement profilealong the beam and the shear link. As shown in the figure,the displacement profile, during the push cycle, between C5and C2 is almost a straight-line indicating that the sheardeformation is dominant. However, the changes in slopebetween C2 and C1 and C5 and C6 indicate additionalbending deformation at these locations as discussed earlier.

A response parameter of greater interest than the totalrotation γ is the link plastic rotation angle γp. The value γp

was computed for each specimen as follows (Engelhardtand Popov, 1989):

where Vlink = shear force in the link ke = ratio of Vlink /γ in the elastic range For each specimen, ke was estimated from the measured

values of Vlink and γ in the initial elastic cycles. The valueof γp computed in this manner includes contributions ofinelastic link deformation as well as inelastic link rotationproduced by yielding outside the shear link. Figures 19 and20 show the M-γp of the BU30 and BU16. The plastic rota-tion of the BU30 specimen reached 8 percent radians wherethe BU16 specimen reached 9.2 percent radians.

CONCLUSIONS

This paper summarized the results of an experimental inves-tigation of two built-up shear links. As expected, the sheardominated the behavior of the two built-up links. The plas-tic rotational capacity for the two links exceeded the codespecified values for such configurations. Based on thesetests the following conclusions and observations may bederived:


Fig. 17. Rotation link at inflection point.Fig. 18. Deflection profile along BU30 length.

1 6

L

δ − δγ =

linkp

e

V

kγ = γ −

Fig. 19. M-γp Response of BU30. Fig. 20. M-γp Response of BU16.

• Built-up shear links designed according to the 1997AISC Specifications achieved the desired behavior spec-ified in the specification. However, the average over-strength factor for BU30 and BU16 was equal to 2.1,which is almost 31 percent more than the AISC specifiedvalue. Therefore, caution should be used in utilizing theAISC overstrength ratio for built-up shear links. Addi-tional experimental investigation is needed to propose anoverstrength factor for built-up shear links.

• The brittle facture in the bottom flange of BU30 and hairline crack in weld area of BU16 showed that the eccen-tric brace connection played an important role in thecyclic behavior of the built-up shear link.

• Both specimens exhibited good ductile behavior and didnot show any sign of local buckling in the link web andflanges or lateral-torsional buckling. This indicates thatthe design equations in the AISC Seismic Provisions canbe used for the design of built-up shear links.

NOTATIONS

a = Stiffener spacingb = Flange widthd = Total depthtf = Flange thicknesstw = Web thicknessFya = Actual yield stress based on coupon testingFy = Minmum specified yield stressFye = Expected yield stressRy = Ratio of expected yield stress Fye to minimum

specified yield stressD = Web DepthMp = Nominal plastic flexural strengthVp = Nominal shear strength of active linkVpe = Expected shear strength of active linkVpa = Shear strength based on Fya

Vpm = Measured yield strength on the onset of webyielding

Vu = Ultimate web shear capacity measured duringexperiment

δy = Measured displacement when yielding initiatedin the web panels

Ωa = Ratio of Vu to Vpa

Ωs = Ratio of Vu to Vp

Ωe = Ratio of Vu to Vpe

Ωm = Ratio of Vu to Vpm

δ1 = Measured displacementδ2 = Measured displacementL = Distance between center line of the actuator to

face of continuity plateγ = Total link rotationγp = Plastic link rotationke = Ratio of measured shear force to g in the initial

elastic cycles

ACKNOWLEDGMENTS

The authors would like to thank the California Departmentof Transportation for sponsoring the work reported in thispaper. Special thanks for Mr. T. Leahy and Mr. A. Akin-sanya and other Caltrans engineers for the help and cooper-ation during this Project. Professor E. Popov comments andadvice during the design and the testing phases of the spec-imens are sincerely appreciated. The cooperation and helpfrom the Gerwick, Severdrup, DMLM Joint Venture is alsoappreciated. The help of Ms. E. Ware in preparing some ofthe figures is appreciated.

REFERENCES

American Institute of Steel Construction, Inc. (AISC)(1997), Seismic Provisions for Structural Steel Buildings,April 15, Chicago, Illinois.

Malley, J.O. and Popov, E.P. (1983), “Design Considera-tions for Shear Links in Eccentrically Braced Frames,”Report No. UCB/EERC-83/24, Earthquake EngineeringResearch Center, University of California, Berkeley.

Kasai, K. and Popov, E.P. (1986), “A Study of SeismicallyResistant Eccentrically Braced Frames,” Report No.UCB/EERC-86/01, Earthquake Engineering ResearchCenter, University of California, Berkeley.

Engelhardt, M.D. and Popov, E.P. (1989), “Behavior ofLong Links in Eccentrically Braced Frames,” Report No.UCB/EERC-89/01, Earthquake Engineering ResearchCenter, University of California, Berkeley.

Hjelmstad, K.D. and Popov, E.P. (1983), “Seismic Behaviorof Active Beam Links in Eccentrically Braced Frames,”Report No. UCB/EERC-83/24, Earthquake EngineeringResearch Center, University of California, Berkeley.

California Department of Transportation (Caltrans) (1992),“Standard Specifications,” Sacramento, California.

Applied Technology Council (ATC) (1992), Guidelines forCyclic Seismic Testing of Components of Steel Structures,ATC-24, Redwood City, California.

Vincent, J. (1996), “Seismic Retrofit of the Richmond-SanRafael Bridge,” 2nd US Seminar on Seismic Design,Evaluation and Retrofit of Steel Bridges, Report No.UCB/CEE-Steel-96-09, University of California, Berke-ley.

Itani, Ahmad (1997), “Cyclic Behavior of Richmond-SanRafael Tower Links,” Report No. CCEER 97-4, Centerfor Civil Engineering Earthquake Research, University ofNevada, Reno.

American Association of State Highway and TransportationOfficials (1996), Standard Specifications for HighwayBridges, Washington, D.C.

Hamburger, R. and Frank, K. (1994), “Performance ofWelded Steel Moment Connections, Issues Related toMaterials and Mechanical Properties,” Invitational Work-shop on Steel Seismic Issues, SAC Joint Venture, Sept. 8and 9.


Behavior of Built-Up Shear Links Under Large Cyclic Displacement

Documents

Transcript of Behavior of Built-Up Shear Links Under Large Cyclic Displacement